Gene Expression Correlates of Language Regression in Autism Spectrum Disorder

Background: Language regression occurs in approximately one-third of children with autism spectrum disorder (ASD) in the first three years of life (Goin-Kochel et al., 2014). Individuals with ASD who experience regression were found to have poorer outcomes of adaptive, cognitive, and social-communicative functioning, compared to those with ASD with no regression (Parr et al., 2011). However, the mechanisms underlying this behavioral phenotype remain unknown (Jones & Campbell, 2010). Recently, gene coexpression network analysis has identified specific spatial and temporal expression patterns in genes associated with ASD (Willsey et al., 2013). Further analysis of gene expression trajectories in the canonical language region of the superior temporal cortex during prenatal and early postnatal development may help to reveal molecular mechanisms of language regression in children with ASD.

Objectives: To investigate superior temporal cortical gene expression correlates of language development and regression in individuals with ASD with likely gene disrupting mutations.

Methods: Participants were 353 children from the Simons Simplex Collection with likely gene disrupting mutations (as defined by Iossifov et al., 2014) who meet strict criteria for ASD. Superior temporal cortical gene expression data from eight weeks post-conception to three years post-birth was extracted from the BrainSpan transcriptome RNA-seq data (http://www.brainspan.org). Coexpression modules were constructed using signed weighted gene coexpression network analysis (WGCNA; Parikshak et al., 2013). 56 children with parentally reported history of developmental regression in language skills were compared to 294 children without language regression. First, a nonparametric analysis was performed to examine the relationship between the presence of language regression and gene expression trajectories as identified in coexpression modules. Next, coexpression groups were compared using one-way analysis of variance on parentally reported age of first spoken single words and age of first spoken phrases.

Results: Five coexpression modules were identified with two modules representing clear increases in expression levels and one module representing decreasing levels of expression over prenatal and early postnatal development. No relationship was found between coexpression module and rate of language regression (χ2 = 4.44, p= 0.35). In addition, no significant differences in age of first spoken words (F(4,289) = .542, p = .71) or first spoken phrases (F(4,270) = 1.001, p= .41) were found among coexpression module groups.

Conclusions: Gene coexpression network analysis was used to identify differential gene expression patterns among children with ASD with and without language regression. Our evidence suggests that differential gene expression trajectories in the superior temporal cortex are unlikely to contribute independently to the language regression seen in some children with ASD. Future studies should investigate gene coexpression networks focused on other brain regions implicated in language development. Better understanding of the developmental timing of gene expression in genes disrupted in individuals with ASD may aid in explaining the phenotypic variability in language outcomes among children with ASD.